Wind profile prediction using a Meta-cognitive Fully Complex-valued neural network

Abstract

This paper applies the recently developed Meta-cognitive Fully Complex-valued Radial Basis Function (Mc-FCRBF) network for predicting the speed and direction of wind. Mc-FCRBF network contains two components: a cognitive component and a meta-cognitive component. A Fully Complex-valued Radial Basis Function (FC-RBF) network is the cognitive component and a self-regulatory learning mechanism is its meta-cognitive component. In each epoch of the training, when the sample is presented to the Mc-FCRBF network, the meta-cognitive component decides what to learn, when to learn, and how to learn based on the knowledge acquired by the FC-RBF network and the new information contained in the sample. Performance comparison of the meta-cognitive fully complex-valued RBF network (Mc-FCRBF) applied for wind speed prediction shows better prediction of wind profile (Speed) characteristics when compared to a real-valued extreme learning machine and FC-RBF network.